Five years ago, we established a CRADA with Boehringer-Ingelheim Pharmaceuticals to generate a panel of mAbs targeting expanded human Tregs (hTregs). Our initial strategy was to immunize mice with hTregs expanded in culture for 14 days, and then screen the resulting hybridomas for binding to expanded hTregs, but not to activated Tconv cells. We subsequently altered our strategy to minimize responses to non-Treg antigens by immunizing first with HeLa cells, followed by cyclophosphamide treatment to deplete responding B cells. Supernatants from hybridomas generated by this protocol were screened by flow cytometry for preferential binding to expanded Treg cells compared to expanded Teff cells and HeLa cells. Hybridomas were also tested in vitro in T cell suppression assays to look for modulation of Treg function. mAbs were also used in western blot analysis to identify target antigen sizes. mAbs that showed binding in western blots, were also tested for their ability to immunoprecipitate their target antigen. We grouped mAbs together based on staining patterns, antigen size as determined by western blot analysis or as unique antibodies. One large group of mAbs identified an epitope on CD25 that was expressed at high levels on expanded Treg, but at much lower levels on expanded Teff cells. These unique anti-CD25 mAbs did not inhibit T cell proliferation, modulate Treg suppression, or block IL-2 signaling (proliferation or STAT5 phosphorylation). We currently have 12 additional candidate mAbs that require further characterization to identify their target antigens. These mAbs have failed to reproducibly bind antigen in western blot analysis and we have been unable to immunoprecipitate their target antigens for mass spec analysis. To circumvent this difficulty, we submitted these antibodies for a surface receptor microarray screen carried out by Retrogenix. Retrogenix has a library of over 4500 human membrane proteins individually over-expressed in human cells. Most of the target antigens of these mAbs have now been identified. To determine whether a similar selective binding could be demonstrated in vivo, we used the xeno-GVHD model in which NOD-SCID gamma common-chain deficient (NSG) mice are engrafted with human PBMCs. Several of the anti-CD25 mAbs bound to CD25 expressed by Tregs activated in vivo, but not to CD25 expressed on activated Teffs. In contrast, both Daclizumab and Basiliximab, two anti-CD25 mAbs used clinically for treatment of multiple sclerosis and graft rejection, reacted equally to CD25 expressed on in vivo activated Tregs and Teffs. These studies suggest that the unique anti-CD25 mABS we have generated may prove to be useful to selectively deplete Tregs from the tumor microenvironment.